Live data from Hacker News

M1 Pro First Impressions: Core Management and CPU Performance

eclecticlight.co

41–50 of 234 posts

Re: M1 Pro First Impressions: Core Management and CPU Performance

#41

I can’t wait for the desktop-class Pro hardware. If you extrapolate on power consumption alone it would be a complete monster. They could do something gnarly like fit 8 sockets on a single board and send your compute ability to level 9000. I’m hoping for a mid-tier unit though - a rebirth of the G4 Cube as a “Mac Mini Pro” I daily drive an M1 mini now and abuse the hell out of it. I’ve never heard the fan. It’s a rem…

> They could do something gnarly like fit 8 sockets on a single board

Those sockets need to communicate with each other, and extremely quickly. And that I/O is very, very expensive in terms of power. Even with that you still end up with poor performance scaling, especially with anything that isn't numa aware (which almost nothing is)

So no, that extrapolation doesn't make sense. But even ignoring that missing cost, 8x M1 Max's would be ~800w, give or take. That would work in a Mac Pro replacement, but that'd still be a significant power increase over many of the existing models.

Re: M1 Pro First Impressions: Core Management and CPU Performance

#42
post #36

Earlier quoted context omitted.

And if its anything like my regular M1, it does that while remaining cold. I used a 2019 macbook pro and the thing would be roasting hot if you had docker running.

The regular M1 uses far less peak power than the M1 Pro/Max. They’re definitely still far more efficient than competitors though.

Can you overclock the regular m1 then?

Re: M1 Pro First Impressions: Core Management and CPU Performance

#43

Earlier quoted context omitted.

I've seen little to no hype based on synthetic benchmarks. M1 macs are just really damn fast for daily usage. My M1 Macbook Air blows my 16" out of the water in every conceivable way. I'd elaborate but there seriously isn't any dimension of it that isn't an improvement. Maybe the smaller display?

> My M1 Macbook Air blows my 16" That is the issue though. Intel macbooks were notoriously badly cooled. That hurts the performance a lot. If all you have used is that, of course you find the new one a lot better. However, compared to any other high end Intel/AMD laptop, you would not perceive this much difference in speed. Pick up a Macbook 16 and XPS 15 from 2019 with the same specs (CPU/RAM) and the XPS absolutely…

And weirdly one of the major design changes for the new M1 Pro & Max machines were significantly improved cooling. Weird how suddenly Apple cared about airflow when it was their 100w SoC in the machine instead of Intel's....

Re: M1 Pro First Impressions: Core Management and CPU Performance

#44
post #10
post #3

"Surprisingly, these benchmarks seldom exceed 50% load on any of the ten cores, which raises the question of how accurately they represent maximum CPU performance." Eh, maybe. I'm not inclined to rely on the reported % CPU usage to represent anything useful. For example, it's very easy to be at 100% CPU when you're actually spending most of that time waiting for memory reads.

The way CPU usage is measured includes time waiting for memory reads. Only a hyperthreaded core would be able to schedule a different task on such a core. For the M1 Pro, which is not hyperthreaded, if all dispatched instructions are waiting for memory reads, then the retired instruction counter simply does not get incremented for that cycle.

Terminology nitpick: SMT, not hyperthreading. Hyperthreading is an Intel trademarked marketing term for their implementation of SMT, like vPro is their term for platform management, etc.

Re: M1 Pro First Impressions: Core Management and CPU Performance

#45
post #12

Mine just arrived today and I'm blown away. After upgrading our app's dependency chain to run native arm64 builds (which took a bit of googling) webpack's incremental build time on our app is down to 116ms from 983ms on my 2016 Macbook Pro. Over 8x faster. Our Tensorflow.js (webgl) models that previously ran about 8fps in the browser are now running at the camera's native 30fps framerate. (Similar to my desktop 2080t…

> After upgrading our app's dependency chain to run native arm64 builds

Would you mind sharing your research results?

I am getting one soon and I am in progress of drafting notes for stuff like migration & setup.

Re: M1 Pro First Impressions: Core Management and CPU Performance

#46

These laptops are too expensive. What the hell is the markup on these things

I believe these are actually one of the lowest markups Apple has and in general a great deal. Consider this: 1. 14” or 16” custom miniLED display 2. Anodized one-piece aluminum clamshell 3. Completely custom SoC (much more expensive than buying Intel’s latest, at least in the short-term) 4. Completely custom OS with a full suite of native apps 5. Custom MagSafe adapter 6. Thunderbolt 4 (3 of them) Now consider that t…

They must be getting their bread by charging $400 for the ram upgrade instead of the traditional $200 apple price for what should be $80 of memory.

Re: M1 Pro First Impressions: Core Management and CPU Performance

#47
post #33

I am curious, since Intel is relying more and more on P and E cores as well, is there any reference or research available for optimizing multithreaded userland process tasks with varying QoS? A lot of the pthreads books I see are from the late 90s. Is there a more recent reference? What's the best way to write cross-platform (e.g. not Grand Central Dispatch) multithreaded apps with these new chip architectures?

I just ran into this tweet for Intel: https://twitter.com/DeepSchneider/status/1456314755380097027 They move everything that isn't foreground to an efficiency core, which is awful for compiling or video processing. There's apparently a BIOS option that will use ScrollLock for disabling the efficiency cores entirely.

Thank you for sharing this, it's interesting - I've also gotten the impression (but lack citation) that Intel E cores are targeted at thermal isolation instead of power minimization as the M1 may target.

This is front of mind for me since reading a Cloudflare blog regarding AVX-512 instructions invoking dynamic frequency scaling to manage power/thermal capacity on chip. (https://blog.cloudflare.com/on-the-dangers-of-intels-frequen...)

If this is happening on Xeons, it's probably happening on consumer dies as well, in addition to other non-obvious power/performance optimizations. Perhaps this is why Alder Lake is pumping up the TDP[1]?

edit: [1] https://news.ycombinator.com/item?id=29106860

Re: M1 Pro First Impressions: Core Management and CPU Performance

#48

Earlier quoted context omitted.

> My M1 Macbook Air blows my 16" That is the issue though. Intel macbooks were notoriously badly cooled. That hurts the performance a lot. If all you have used is that, of course you find the new one a lot better. However, compared to any other high end Intel/AMD laptop, you would not perceive this much difference in speed. Pick up a Macbook 16 and XPS 15 from 2019 with the same specs (CPU/RAM) and the XPS absolutely…

And weirdly one of the major design changes for the new M1 Pro & Max machines were significantly improved cooling. Weird how suddenly Apple cared about airflow when it was their 100w SoC in the machine instead of Intel's....

Totally agree. And so many of the changes that people are enjoying about the new MacBooks (namely reverting Apple's failed experiments like TouchBar, only USB-C ports, and the butterfly keyboard) were all introduced alongside Apple Silicon. Not saying I blame them — it makes a ton of sense from a business/product marketing perspective, especially bundling the new display too.

Re: M1 Pro First Impressions: Core Management and CPU Performance

#49
post #46

Earlier quoted context omitted.

I believe these are actually one of the lowest markups Apple has and in general a great deal. Consider this: 1. 14” or 16” custom miniLED display 2. Anodized one-piece aluminum clamshell 3. Completely custom SoC (much more expensive than buying Intel’s latest, at least in the short-term) 4. Completely custom OS with a full suite of native apps 5. Custom MagSafe adapter 6. Thunderbolt 4 (3 of them) Now consider that t…

They must be getting their bread by charging $400 for the ram upgrade instead of the traditional $200 apple price for what should be $80 of memory.

> traditional $200 apple price

That might've been for when it was only an 8GB increase in RAM.

Post reply on HN